It is often found that a large proportion of failures in a product are due to a small number of causes. By use of Pareto analysis many problems which fall in the above category can be solved. Therefore, if we analyze the failure data, we can determine how to solve the largest proportion of the population that belongs to few significant causes of failures. However, before committing resources it is important to make sure that the data have been fully analyzed, to obtain the maximum amount of information contained therein. A Pareto plot is often useful in analyzing the failure data. 5.5.1. Pareto Analysis A batch of C-band Upconverters were subjected to 48 hours of ESS. The units are vapour sealed because they are designed to work in the outdoors. The ESS constitutes of only hot and cold testing at -400C to +600C with two ours at each extreme. The units are powered on and tested dynamically. Fig. 1 shows failure data on a C-band microwave Upconverter taken from production records. These data indicate that attention paid to burning of 35V tantalum capacitors, soldering of eyelet to the pads of the plated through holes, improving the quality of shoulder washer used for the regulator connection to heatsink would be likely to show the greatest payoff in warranty/post production cost reduction. The corrective action taken for the above four failure modes resulted in significant improvement in reliability and warranty failures are bound to reduce correspondingly on the future production.
| Sl.NO. | Failure description | % failued |
|---|---|---|
| 1 | 10uF, 35 V Tantalum capacitors blowing off | 13 |
| 2 | eyelets showing intermittant contat | 9 |
| 3 | Shorting of teflon shoulder washer of a regulator | 8 |
| 4 | Thermister failure | 3 |
| 5 | RF absorber falling off C-band amplifier cover | 2 |
| 6 | Oscillator/Multiplier failure due to VCC short | 2 |
| 7 | Miscellaneous | 10 |
Other failure modes should also be considered, since, corrective action may be relatively simple and therefore worthwhile. For example, falling off an RF absorber that was glued to the cover plate of microwave amplifier was found to happen in very few cases. However, it was easily fixed by roughening up the surface and using a primer before using epoxy to fix the absorber.
The failure data can also be analyzed in relation to contributions to down-time or by repair cost, depending upon the criteria of importance. The data must be analyzed to reveal as much as possible about the relative severity of problems and the likely causes. In addition adequate number of data as the basis for decision making on where to apply effort to improve reliability is required.